openModelleropenModellerframework for ecological niche modellingframework for ecological niche modelling
CRIA, INPE, Poli-USP
Why openModeller?
Proof of concept: specimen occurrence data as infrastructure for research and policy making
First prototype developed as part of the speciesLink project (2001-2005)
Open source, C++, available at sourceforge
Attracted international collaboration: University of Kansas
and University of Reading
Why openModeller?
Modeling process is usually quite complex and time consuming
requires great expertise in a number of software and tools (different formats, different software, …)
framework capable of dealing with different projections, coordinate systems, and formats
By using such a framework, specialists would be able to concentrate more on the analysis then on preparation of data
Thematic project funded by Fapesp (2006 – 2009)
Occurrence
data
Distribution model
(environmental space)
Tem
pera
ture
Precipitation
Environmental
layers
algorithm
openModeller basic functionsopenModeller basic functions
Work with multiple algorithms.
Allow different interfaces on top of the same library
(desktop,
command line, web)
Support different input formats (for both points and
rasters).
Platform independent (runs on GNU/Linux, Mac OSX and
Windows).
local files GBIF local files
Occurrence data (points) Environmental data (rasters)
openModeller
Algorithm 1
Algorithm 2
Algorithm n
Interfaces (desktop, command line, web service, etc.)
data cleaning
georeferencing
modelling, pre-analysis, post-analysis API
occurrence
data
reader
environmental
data
reader
GEOSSTerraLib
openModeller architectureopenModeller architecture
plugins
speciesLink
Reference Center on Environmental InformationReference Center on Environmental Information
Brazilian not-for-profit, non-governmental organization that aims to contribute towards a more sustainable use of Brazil's biodiversity through dissemination of high quality information.
www.cria.org.br
The The speciesspeciesLink network (splink.cria.org.br)Link network (splink.cria.org.br)
speciesspeciesLink evolutionLink evolution
FAPESP Nov 2005
~700k records
~40 collections
Out 2008
~3M records
~ 160 collections and
subcollections
45% georef
speciesspeciesLink regional serversLink regional servers
+ coleções de países amazônicos
+ repatriação de dados NYBG, MOBOT, ...
speciesspeciesLink indicatorsLink indicators
66% Plants
28% Animals
5% “broad scope”
1% microorganisms 85% from Brazilian collections
15% from foreign collections
Good models depend on data(availability, usability, quantity and
quality), tools, and Expert knowledge
SP
SC
PR
RS
Expected distribution of
Hypsiboas bischoffi
Potential distribution model of Hypsiboas bischoffi
Thank youThank you
http:// openmodeller . sf . net